Minimizing Error and Bias in Death Investigations

Abstract

One of the prominent developments in the forensic sciences is the emergence of attention to cognitive aspects of forensic examination. Notable in this regard is the recognition that forensic results can be swayed by the examiner’s exposure to non-scientific background information that should arguably have no bearing on the result. To counter these effects, forensic agencies have introduced context management procedures, which are designed to withhold background information from the examiner during critical parts of the examination. Context management procedures are well-suited for some forensic disciplines but apply less obviously to disciplines that entail complex, sprawling, iterative, and open-ended reasoning processes. Notably, the procedures have been met with stern resistance from the field of death investigation. This Article sets out to explore whether and how context management can and should be implemented to the practice of death investigation.

As currently practiced, the death investigation environment is replete with background information that renders investigative conclusions susceptible to influences borne by non-medical information of unknown reliability. Such effects can occur through two routes: either by way of unconscious biasing of the investigation (the context bias) or through the conscious and deliberate incorporation of that information. Both effects could be mitigated by means of context management procedures. To complicate matters, however, background information also plays an important facilitative role in death examinations by way of enabling the generation of investigative hypothesis without which the process is unlikely to succeed. These conflicting effects of background information make it difficult to determine ex ante which information should be shared with the investigator and which should be masked. The Article proposes a nuanced three-part cost-benefit analysis, but finds that it produces only a series of murky judgments that result in a conundrum that defies a simple or uniform solution.

The Article seeks a way out of this conundrum. The framework proposes that referred cases be triaged to identify the small category of ambiguous criminal cases, and that only those cases should be subjected to context management. Within that category of cases, death examiners should be free to receive the types of background information that tend to be within or proximate to their expertise, but not be exposed to types that fall outside of that range. The proposed framework is admittedly imperfect, but given the intractability of the topic, any step forward should be welcomed.